It is known that, for distances of a few metres, the use of a single fibre allows transmission efficiencies of over 80%. Those with step index refraction distribution, e.g. with quartz core and plastic mantle, having a core diameter between 100 and 1000 .mu.m seem to be amongst the most suitable fibres.
However, some difficulties can still arise in coupling these fibres to some kinds of source in which the laser beam is liable to some unsteadiness, and therefore to a slight drift regarding the initial direction that allowed the optimal coupling to the fibre. As a consequence, one does not only have a lowering of the efficiency in the coupling itself, but sometimes one also has damage of the mantle of the fibre and of the mechanical supporting parts due to that part of the laser beam that can no longer enter the quartz core.
In U.S. Pat. No. 3,843,865 (G. Nath) the use of a fibre with the variable section core and with an unconventional mantle (air or liquid of appropriate refraction index) is described. Said fibre allows overcoming the aforesaid drawback and, at the same time, to reduce to a minimum the increase of the angular divergence of the output laser beam. It is also possible to obtain a focal spot of reduced sizes by laying a lens at the output of the fibre.
Besides the abovementioned advantages, Nath's fibre also presents some remarkable drawbacks. Firstly, the methods of core manufacturing seem rather complex and expensive, moreover they limit, at least in practice, to 2-3 m the length of the fibre piece which can be performed. As a matter of fact, the thickening of the fibre terminal makes it difficult to wind it onto a drum besides, the chemical bath method presents some encumberment problems, also. But the most notable drawback of Nath's fibre is the lack of a mantle of traditional type in glass or plastic, in so doing the fibre core turns out to be brittle. The solutions suggested for overcoming the absence of the mantle can most likely present some advantages, e.g. for cooling of the fibre in case of transmitted high powers, but, however, they are complicated and expensive. Furthermore, it seems clear that, at least for powers of 100 W and over, the usual quartz-plastic fibres of large diameter (e.g. 600 .mu.m) are enough.
Besides, either owing to changed requirements of use or in case the output terminal turns out to be damaged, one will have to replace the whole device comprising the variable section fibre. The likelihood of a similar drawback is not negligible also in presence of the flow of the protection gas, because of the particles and fumes emitted by the target.